Planetary gear whine under load 6

[whyeye]

thread406-295989
Hello all, this is my first post but I have loved reading all the technical discussions on this forum for quite a while. We have run into into a potentially show-stopper situation involving the planetary reducer in a new product.

We are dealing with very low power levels by transmission standards, but low noise is an absolute requirement. The input pinion rotates up to 2000RPM with a load of no greater than 0.2Nm. The gears are metric spur gears with 0.75 modulus. Due to cost and manufacturing considerations, helical gears are out of the question.

Here's the problem: under no load, the reducer is virtually silent. This leads me to believe there are no fundamental issues with backlash or center distances etc. However, as soon as any load is applied to the output shaft of the planet carrier, a very noticeable whine becomes present. The noise is not of a whirring nature but is constantly tonal.

Other than profile error, what else could cause an otherwise very quiet gearbox to exhibit so much whine as soon as load is added? I am not a gear designer by profession so am having a very hard time trying to pinpoint the issue. Any help would be greatly appreciated!

 

[Tmoose]

deflection in the housing and support structures?
Is the output load radial or pure torque?

The bearing manufacturers FAG and SKF go into some detail about the importance of establishing ElastoHyrdoDynamic lubrication in rolling bearings, and the requirements of lube viscosity in achieving it.

http://mountingmanager.schaeffler.com/library/library.pdf.wl81.115.e.pdf

http://www.skf.com/group/products/b...ciples/lubrication/viscosity-ratio/index.html

Kappa or viscosity ratio is a measure of whether EHD can exist

Gear lubrication has a strong EHD component, so I'd be looking at my Kappa.

 

[gearcutter]

A lot of your question has already been answered in the thread that you've provided a link to.
If you're asking for specific reasons as to why your gears are generating noise; please provide more info. Supplying drawings that also shows the gear data along with any load conditions data is always a good start.

 

[whyeye]

Thanks for the replies. Radial load would be insignificant in our application so the output can be considered to be pure torque.

gearcutter - I guess the difference to the other thread is that this gearbox is very, very quiet when running under no load. The noise only appears once load is applied, even light load. I have attached the relevant gear data and a 3D view of the gearset. The original data was not in English and I've tried to translate it as best as I could. Hope it's OK if some of the nomenclature isn't AGMA.

Due to the low noise requirement, all gears are made of plastic. The sun and ring gears are PA6/6, planet gears are acetal.

 

[whyeye]

Not sure how to upload multiple files in one post - here is the gear data.

 

[gearcutter]

I've run some quick numbers and the set appears to be well designed to handle the required speed & torque. Based on a service life of 1000 hours and no lube; flank safeties for the sun and planets are a little low but may be acceptable for your application.

The geometry, however, needs optimising.
I'm getting undercut & non-involute contact warnings from the software that I'm using.
This may well explain the noise issue that you're having. The interference will only get worse under load.

Are you using lube, if so what type?
What is the operating environment's temperature?
What are the operating environment conditions.....clean, dusty, etc?
What are the tip diameters of all the components?
What are the base tangent length spans for the sun & planets?
What is the measurement between pins for the annulus?

 

[tbuelna]

Typical "gear whine" is primarily caused by transmission errors. The gear structures vibrate in response to accumulation and release of strain energy. There are many causes of transmission error, but they all result in an interference condition between multiple gear teeth sharing a common load path. Looking at the sketch you provided I would speculate that you might be experiencing excessive deflections in the cantilevered planet pins and cantilevered carrier. The problem is made worse with the use of four planet gears and the fairly high L/D of the gears. The planet carrier structure may be twisting under higher load, which is why you don't detect any noise when rotating the gear set with minimal load.

Adding some face crown to the planet gears would help a bit, but it would also reduce the load capacity of the gears. And since they are made of plastic this might not be practical. The only other thing I might suggest would be to redesign the carrier structure so that it is supported at both ends, and also so that the planet pins are supported at both ends. This would reduce the deflections in the carrier structure due to torsional wind-up, and would maintain much more accurate alignment of the planet gears.

Hope that helps.
Terry

 

[MikeHalloran]

I'm guessing the planet carrier is plastic, and has metal pins pressed or molded in.

If that's true, you might be able to reduce the pin deflection that tbuelna conjectured, by increasing the axial thickness of the planet carrier.
E.g., doubling the thickness of the planet carrier in the 'back' direction and using correspondingly longer pins would certainly help.

If you are painted into a corner already, you might be able to help the situation a little by adding a boss around each pin, extending in the 'forward' direction nearly to the centerplane of the planets, and a corresponding counterbore in the planets. ... balanced of course against changes in the various bearing pressures.

A tapered boss would be even better for supporting the pins, and a corresponding tapered counterbore would retain most of the planets' strength.





Mike Halloran
Pembroke Pines, FL, USA

 

[whyeye]

OK, actual service life only needs to be about 100 hours under typical conditions. Operation is intermittent, definitely no more than 10 minutes per day. We use a fully synthetic lubricating grease. The product is fully enclosed and designed to be maintenance free for its entire service life.

I'm not familiar with some of the terms - what is tip diameter and base tangent length span?

The pin positioning is highly accurate with a measured tolerance of only 0.01mm. Since we are dealing with only a very small amount of torque, I can't imagine that deflection of the carrier structure can be a significant factor. The carrier is made using powder metallurgy so should easily withstand the rated load and hold the pins in place. The pins are 4mm in diameter by 16mm length, 6mm of which is planted into the carrier plate.

I may have miscommunicated the noise issue. The noise is only apparent when the gearset comes into contact with a resonating surface such as a wooden desk. Just holding it in the hand, noise is still very quiet when load is applied.

 

[tbuelna]

Don't be so quick to dismiss the issue of structural deflections in you carrier structure. It would only take a couple thou inch of displacement along the mesh contact to create significant transmission error. And with the cantilevered carrier design you have, I can easily see this being a problem. You should remember that the radial forces acting through the planet pin centers can be quite high. And since the point the planet radial forces are applied has a large axial offset from where the carrier is supported, you should expect a fairly significant degree of relative deflection at the unsupported end of the carrier.

I would recommend that you do a study of the combined clearances/misalignments in the bearing supports of the carrier, planet pins and sun gear. If there is more than just .002" or so of misalignment along the mesh line, then you likely have a problem.

As I noted above, the cantilevered design of your planet carrier is not a good choice. With the small size of the gears you are using it only takes a tiny amount of misalignment at the mesh line to cause problems.

 

[whyeye]

Regarding potential deflections in the carrier structure, it is more a case of my mind thinking "gee I HOPE that's not the case". With production tooling completed for all the components, such an overhaul would be a rather significant undertaking. That being said, self-consolation won't solve the issue at hand. I'm interested to learn how this can be verified through testing. You mentioned a misalignment study - how is this conducted? Are special tools required?

I'm not too sure how the number of planet gears can have an impact on TE? The number of planets is governed by the teeth numbers on the sun and ring gear. In our case, 2 or 4 gears is needed for even spacing.

Is it unrealistic to think that this problem might be solved with some simple tweak?

 

[MikeHalloran]

If you know the actual properties of the P/M parts, you should be able to analyze the stiffness of the planet pins. Manual methods on paper should get you close enough to illuminate the problem.

Or you can just go ahead and measure it, with a dial indicator and some deadweights. Since P/M properties are subject to some variation, direct measurement may be preferred.



Mike Halloran
Pembroke Pines, FL, USA

 

[Strong]

Whyeye,

There are three noise control methods that can be very effective without redesigning the gear set:
1) Constrained layer damping for the ring gear (the usual suspect for your type of problem)
2) Vibration isolator material/device between ring gear and rigid stationary support (eliminates the "hard desk" effect you mentioned)
3) Torsional vibration damper on high speed shaft

These NC methods can be relatively inexpensive, and only one or two methods may be necessary. Typically the sound and vibrations are measured for a baseline and to identify the sound/vibration properties for selecting the appropriate noise control method and materials. Spur gears are usually the noisiest type of gear, so now you have to control it! A reduction of 10-dB (1/2 as loud) or perhaps more should be possible.

Walt
Sound-Vibration Control Engineer

 

[whyeye]

Strong,

This is the first time I've heard of constrained layer damping. Looks very interesting! We may not be able to implement this concept in the current product due to lack of space but will definitely keep it in mind for the future.

The other two methods are in fact already a part of the design, although that is not apparent in the sketch. They are part of the overall system designed to minimize transmitted noise. These measures have certainly been very effective for us in the past. For the present product, it seemed the base vibration is too great to be effectively dampened.

Fortunately, the noise problem has improved significantly after we reassembled all the components. I'm not even sure what we did wrong the first time around. However, it has been noted that turning the planet gears upside down has a noticeable impact. I'm guessing this has something to do with draft angles on the mold.

 

[Strong]

Whyeye,

Can you fit constrained layer damping about 3 mm thick, or perhaps less, on outside of ring gear?

Walt

 

[BrianE22]

As molded gear teeth? Without machining, I wouldn't think you could get the accuracy you need to get a smooth, quiet mesh.

 

[mfgenggear]

I with Brian. To my knowledge what is be asked of plastic gears may be much.
i don't see how the required tolerance could be held and especially after meshing.

 

[Windward]

I thought that plastic gears were the solution to noise.

Why would a plastic helical gear cost more than a plastic spur gear? The helical mold might cost a little more, but after producing a few hundred of each kind, why would there be a price difference?

Understand that I am asking questions, not stating facts. I don't have any facts here.

 

[whyeye]

Strong - Well there is about 0.6mm clearance between the ring gear and outer housing (also cylindrical) which seems sufficient to add a damping layer. I guess whether or not CLD is possible would depend on material suitability. I have read a bit more on the subject and most suggest that you need a stiff (metal?) material for the base and constraining layers. Is this true? Also, how could CLD be designed to hold the torsional forces acting on the ring gear?

Brian - You are right that, with shrinkage, accuracy is not great on our molded gears. I was no aware there were any machining methods for plastic gears post-molding. Can you tell me a bit more on what can be done?

Windward - I agree that once tooling has been completed, part costs for helical molded gears won't be a lot different from spur gears. However, my understanding is we would have to upgrade our current molding equipment due to the more complex demolding process. Whereas with spur gears can be simply ejected via ejector pins, I imagine helical gears would entail some sort of twisting motion. I might be wrong on this matter and would gladly be enlightened.

 

[BrianE22]

I just assumed plastic gears could be cut. It could be that the deflection during cutting is enough that you don't get any more accuracy. I do see in

http://www.thomasnet.com

that there are companies that are listed under "Cut Plastic Gears".